Rockburst is a common geological hazard in underground mining and excavating, constituting a serious threat to the safety of workers and equipment. Rockburst liability prediction is of crucial significance to the prevention of rockburst hazards. Multi-criteria decision-making approach is an effective tool to predict rockbursts. In this paper, five empirical factors (strength-stress ratio, Russenes criterion, brittleness index, strain energy storage index, and integrity index) were used to construct the standard of rockburst liability. A novel analytic hierarchy process (AHP) was proposed to determine the weight of these factors, and then a cloud model was established for rockburst liability prediction on this basis. In the new AHP, the random errors induced by the pairwise comparison matrices of decision-makers were corrected according to the Bayesian theories. Nine cases from engineering projects in China were collected to validate the proposed model for rockburst liability prediction, and the prediction results were consistent with the field status. The quantitative index (Euclidean distance) indicates that the novel analytic hierarchy process has better performance than the original analytic hierarchy process in the weight calculation of factors; the new proposed cloud model is also superior to the original cloud model and the technique for order preference by similarity to an ideal solution model in the prediction of rockburst liability.